How to map a company’s data without erasing its contradictions

Ask three departments in your company how much you sold last month and you will probably get three figures. Sales counts invoiced orders, accounting counts accrued revenue and treasury counts cash collected. Each figure makes sense to the people who use it, and there is rarely a document that says there are three.

That gap shows who decides with which figure and how much work it takes to reconcile them at every close. The data method is built to keep it on the record.

The data method is how XPTNova studies a process, following each piece of data from where it originates to the decision it supports. DataNova, our platform, uses it to build each company’s operating model.

Every business process can be understood, valued and redesigned by following the path of its data.

The unit of the method is the mention and its source

The usual way to document data is the data sheet, which records the name, definition, owner and system of each item. Filling it in requires writing down a single definition.

If sales, accounting and treasury gave three definitions of revenue, whoever fills in the sheet picks one. The other two disappear, and with them the finding that there were three. Anyone who reads the sheet later will not know the disagreement ever existed.

That is why the method records mentions. A mention is each time a source names a piece of data, copied exactly as it appears, with its source and its locator. The locator is the exact place, such as a section of a procedure, a cell in a spreadsheet or a turn in an interview, and it lets anyone go back and check what was said.

Reconciling is decided later and written down separately, with its evidence. When two sources contradict each other, the contradiction is recorded with both sides and nobody resolves it during mapping. Which source prevails is decided by the company, in formal minutes.

Four stages, each with its own gate

Each of the four stages ends at a gate, the list of conditions to meet before moving on. During mapping, for example, a mention without a locator keeps the stage from closing. Whatever the company does not have is declared as a gap, along with its consequence for the analysis.

• Inputs. The starting point is the objective, written as the decision someone will make with the result, and that determines what to request from the company. The filled-in form is always requested, because a blank template shows which fields exist and a completed one shows whether they are used.
• Mapping. Every appearance of a piece of data in documents, systems, spreadsheets, emails and interviews enters as a mention. The map shows where each piece of data originates, who produces it, who consumes it, how it is transformed and which sources contradict each other about it.
• Analysis. The map is asked what it says about the business, for example which data originates in several places or which decision is made with a figure that has no origin. Each candidate finding is checked against real operations.
• From analysis to objective. The output is whatever the objective calls for, whether a system to build, a plan to close compliance gaps or a controls matrix, and every element points back to the row of the map that produced it.

What lives in someone’s head

Much of a mid-sized company’s data sits in no system at all. A customer’s credit limit gets approved over email. Only one person in treasury knows how to do the bank reconciliation, using her own file. The warehouse supervisor works out the reorder point for each item from memory.

Documents tell you which data exists and where it is stored. People tell you where each piece comes from, who uses it and what happens when it fails. On the origin, owner and use of a piece of data, conversation is the source that prevails, so interviews are planned as carefully as the document request.

The figure mentioned in passing is captured too, such as the number of days an analyst says the close takes her. It is recorded with its number, unit and period, marked as stated in an interview, so it can later be checked against a record.

Every figure says where it came from

Every figure in the analysis carries a letter that says where it came from. There are seven, ranging from the general ledger and system records to interviews, assumptions and market data. A ledger balance and a number recalled in an interview do not carry the same weight, and the letter puts that in writing.

A savings figure requires the mechanism that produces it and its inputs, such as volume, effort per case or the cost of rework. When those inputs do not exist, the method proposes measuring the baseline first.

Absence is written down too. The method distinguishes data no source stated, fields that do not apply, needs with no source and sources that were read and did not contain the data.

From the map to the decision

When the objective is automation, the result is a target system with a project plan. The plan starts with an initial version that captures each piece of data at the step where it originates, moves cases out of email, gives every department the same file and measures the process with records.

That version also forces a decision on whatever the map left open, because a form needs to know who owns each piece of data. Automation projects come afterwards, in the order the baseline sets, and every proposed function points back to the data that gave rise to it and to its source.

The method also states its limits. Interviews cover part of the company’s people, and that share is reported. The analysis describes the company as of a given date, and every document carries its cutoff date.